EXPOSURE RESPONSE ANALYSIS TO

EVALUATE A DRUG'S EFFECT ON ECG

PARAMETERS

Christine Garnett, PharmD

DCRP, FDA

April 6, 2016

CSRC/FDA Workshop: The Proarrhythmic

Assessment of New Chemical Entities

Disclosures

– The views expressed in this presentation are that of the author

and do not reflect the official policy of the FDA. No official

endorsement by the FDA is intended nor should be inferred.

– The data presented are publicly available

2

10-year Experience with

Regulatory Decision-Making

2005

• ICH E14

• Analysis of the relationship between drug concentration and QT/QTc interval changes is under active investigation

2008

• FDA publication on role of C-QTc in regulatory decision-making

• Waive TQT for QT prolonging drugs; predict QT effects at doses not studied in TQT study; evaluate assay sensitivity of positive control; and perform dose adjustments in special populations

2014

• ICH E14 Q&A (R2)

• Importance of model pre-specification; provide insight into regimens not studied directly; predict QTc effects of intrinsic & extrinsic factors that affect PK; inform dose selection; clarify ambiguous results

2015

• ICH E14 Q&A (R3)

• Used as an alternative method to the IUT to serve as primary basis for decisions to classify the risk of a drug; data can come from other clinical pharmacology studies than TQT study

3

Outline

• C-QTc analysis for regulatory decisions

• ICH E14 Q&A (R3) key changes to design and analysis

of QT data

• Implementation of ICH E14 Q&A (R3)

– Analysis expectations

– Assay sensitivity

– Difficult drugs

4

C-QTc Provides Clinically Relevant

Perspectives of TQT Results

• Doses not directly

studied

• Specific populations

• Formulation

changes

• Ambiguous results

using IUT

Garnett et al. J Clin Pharmacol. 48(1), 13-8 (2008).

5

Examples where C-QTc Modeling

had Impact on Regulatory Decisions

•Project the QTc prolongation in elderly and renally impaired patients in the product label.

Anzemet (dolansetron)

•No single intravenous dose of ondansetron to exceed 16 mg due to the risk of QT prolongation.

Zofran (ondansetron)

•Project the QTc prolongation at the 40-mg dose, which was not directly evaluated in the TQT study.

Celexa (citalopram)

•Characterize QTc prolongation in the patients from ECGs obtained in phase 3 trials for the product label.

Caprelsa (vandetanib)

•Characterize QTc prolongation in patients with highly variable pharmacokinetics.

•Project QTc prolongation in patients with hepatic impairment.

Ranexa (ranolazine)

•Supported no precautionary labeling statementsEdurant

(rilpivirine)

•QTc prolongation (2-5 ms) is predicted by CEM and not the mean values reported from the IUT analysis (5-10.5 ms)

Saphris (asenapine)

•Project QTc prolongation in patients who are CYP2D6 poor metabolizers for benefit-risk assessment.

Sertindole

Drug information publicly available from FDA website

6

Good Concordance Between C-QTc

and IUT in TQT Studies

Hypothesis:

H0: ≤0

H1: >0

TQT Studies (N=204)

E14 negative

(n=165)

E14 positive

(n=39)

Slope from

C-QTc

relationship

≤086%

8%

(Discrepant)

>0 14%

(Discrepant)92%

Florian et al. J Clin Pharmacol. 2011;51:1152-1162.

Zhang et al. Therapeutic Innovation & Regulatory Science 2015, Vol. 49(3) 392-397

Considerations:• Slope alone does not give full assessment

of QTc prolongation risk

• Drugs can have a positive slope but no clinically meaningful QTc prolongation

7

Linear C-QTc models have adequate

sensitivity and specificity

Garnett et al. Clin Pharmacol Ther. 2016 Mar 4. doi: 10.1002/cpt.361.

8

IQ/CSRC Study: Phase 1 data to

replace the TQT study

• The goal of the study was to

demonstrate that a small dose-

escalation study could detect the

QT effects of drugs known to

prolong the QTc interval around

the regulatory threshold.

• The drugs and doses for this study

were selected by the FDA with an

understanding that successful

detection of the drugs would form

the basis for the use of C-QTc

approach in FIH studies.

Darpo et al. Ann Noninvasive Electrocardiol. 2014 Jan;19(1):70-81. Epub 2013 Dec 30

Each drug was given to nine subjects (six for placebo) in two dose levels

9

Darpo et al Clin Pharmacol Ther. 2015 Apr;97(4):326-35.

Alternative pathway to assess QT

effects of new drugs

10

Drug

Slope,

mean

ms per

ng/mL

LB

90% CI

UB

90% CI

Treat-

ment

effect

ms

Cmax

Day 1,

ng/mL

Projected

QTc

effect

mean, ms

LB

90%

CI*

UB

90%

CI*

Positive drugs (Day 1)

Ondansetron 0.033 0.025 0.042 0.2 284 9.7 6.2 12.8

Quinine 0.004 0.0034 0.0047 -3.0 3623 11.6 6.8 17.1

Dolasetron 0.021 0.013 0.028 3.1 211 7.4 3.0 11.0

Moxifloxacin 0.0065 0.0059 0.0072 2.3 1862 14.5 10.5 17.7

Dofetilide* 22.2 18.9 25.6 1.1 0.42 10.5 6.3 14.9

Negative drug (Day 2)

Levocetirizine 0.0014 -0.0013 0.0041 0.7 1005 2.1 -2.3 6.1

*: Slope from linear model for comparison. Predicted effect for dofetilide using Emax model: 11.6 ms; 90% CI 7.0 to 16.0

C-QTc Relationship as Primary

Analysis

• When using a concentration-response analysis as the primary

basis for decisions to classify the risk of a drug, the upper

bound of the two-sided 90% confidence interval for the QTc

effect of a drug treatment as estimated by exposure-response

analysis should be <10 ms at the highest clinically relevant

exposure to conclude that an expanded ECG safety

evaluation during later stages of drug development is not

needed. (See E14, Section 2.2.4 and Q&A #7)

ICH E14 Q&A (R3)

11

Sources of Data for C-QTc Analysis

• Concentration-response data need not come from a dedicated

QT study

– Data can be acquired from first-in-human studies, multiple-

ascending dose studies, or other studies. Test for heterogeneity.

– Robust, high-quality ECG recording and analysis sufficient to

support a valid assay for ECG intervals

– If there are data characterizing the response at a sufficiently high

multiple of the clinically relevant exposure a separate positive

control would not be necessary.

12

Using Phase 1 Data as Definitive

Assessment to Exclude QTc Prolongation

NonclinicalPhase

1Phase

2Phase

3FDA Filing

• High quality ECGs in SAD/MAD studies

• Intrinsic and extrinsic factors affecting PK

• Clinical doses

KEY DECISIONS

High quality ECGs?

Adequate exposure margin?

QTc data adequately characterized using ER?

Assay sensitivity?

No

Dedicated QT assessment (TQT)

Include additional QTc at high exposure

Yes Use Phase 1 data as definitive assessment

• In vitro and in vivo assays per ICH S7b

13

Assay Sensitivity

• If there are data characterizing the response at a sufficiently

high multiple of the clinically relevant exposure a separate

positive control would not be necessary.

– C-QTc relationship over large exposure range can substitute for

assay sensitivity (no further evaluation)

– Exposure margin is drug dependent, e.g.,

• 3-5-fold higher then exposures observed in patients with renal

impairment or taking metabolic inhibitor

– Challenging for poorly tolerated drugs and drugs with nonlinear

PK

• Moxifloxacin positive control in TQT study

• Non-pharmacological evaluations??

14

Concentration-QTc Model

• Best practices document is

currently being written

• Pre-specified modeling and

analysis plan

• Adequate model fit will have:

– Goodness-of-fit plots show

minimal bias with respect to

concentrations, studies,

time/day, doses

– Good parameter precision Concentrations

15

Potentially Difficult Drugs to Assess

Using C-QTc in Phase 1 Study

• Drugs with heart rate effects

– Inadequate heart rate correction; potential for QT/RR hysteresis

• Drugs with multiple moieties (parent and metabolites) that inhibit

hERG channel

– Single dose studies might not capture effects; C-QTc modeling of

multiple variables is challenging; interpretation of model can be difficult.

• Extended-release formulations

– C-QTc modeling of narrow concentration range can give incorrect

results.

• Drugs with PK/PD hysteresis

– Timing of ECG/PK samples is important for model; PK model needed.

16

Key Messages

• Concentration-QTc analysis can be used as primary analysis in TQT or

Phase 1 studies. This change is based on:

– 10-years experience in applying these analysis to TQT studies

– Good concordance between the results of IUT and concentration-QTc analysis

even though the two approaches do not test the same hypothesis

– Simulations studies (internal and external) evaluating the statistical properties of

concentration-QTc analysis

– Positive IQ/CSRC study results

• C-QTc analysis of Phase 1 data can substitute for a TQT study. Important

considerations include ECG quality, exposure margin, C-QTc model and

assay sensitivity.

• C-QTc relationship over large exposure range can substitute for assay

sensitivity (no further evaluation)

• Not all drugs are good candidates for Phase 1 paradigm. Some drugs will

require special considerations to the study design and/or analysis.

17